I have a concern that after supporting UPDATE/DELETE CASCADE, the
performance would drop.
On Thu, Jul 27, 2017 at 12:54 PM, Alexander Korotkov
wrote:
>
> I wonder how may RI trigger work so fast if it has to do some job besides
> index search with no results?
>
Best Regards,
Mark Rofail
ction.
> If so, we don't introduce additional restriction here...
>
You mean to remove the wrapper query ?
> GROUP BY would also use default btree/hash opclass for element type. It
> doesn't differ from DISTINCT from that point.
>
Then there's no going around this limitation,
Best Regard,
Mark Rofail
This is the query fired upon any UPDATE/DELETE for RI checks:
SELECT 1 FROM ONLY x WHERE pkatt1 = $1 [AND ...] FOR KEY SHARE OF
x
in the case of foreign key arrays, it's wrapped in this query:
SELECT 1 WHERE
(SELECT count(DISTINCT y) FROM unnest($1) y)
= (SELECT count(*) FROM () z)
Th
the experience to answer that
Best Regards,
Mark Rofail
depending on what indexes happen to be available.
>
> Err ...
>
> > I think what you meant to say is that the planner may only choose an
> > optimization of this sort when the index's opclass matches the one
> > DISTINCT will use, ie the default for the data type.
I understand the problem. I am currently researching how to resolve it.
Best Regards,
Mark Rofail
These are limitations of the patch ordered by importance:
✗ presupposes that count(distinct y) has exactly the same notion of
equality that the PK unique index has. In reality, count(distinct) will
fall back to the default btree opclass for the array element type.
- Supported actions:
✔ NO ACTIO
On Fri, Jul 28, 2017 at 1:19 PM, Erik Rijkers wrote:
> One small thing while building docs:
>
> $ cd doc/src/sgml && make html
> osx -wall -wno-unused-param -wno-empty -wfully-tagged -D . -D . -x lower
> postgres.sgml >postgres.xml.tmp
> osx:ref/create_table.sgml:960:100:E: document type does no
of the referencing array or the
whole array. I think there's a lot of options the user might want to
consider and it's hard to generalize to DELETE CASCADE. Maybe new grammar
would be introduced here ?|
Best Regards,
Mark Rofail
g/message-id/28617.1351095...@sss.pgh.pa.us
[2]
https://www.postgresql.org/message-id/CAJvoCutcMEYNFYK8Hdiui-M2y0ZGg%3DBe17fHgQ%3D8nHexZ6ft7w%40mail.gmail.com
[3]
https://www.postgresql.org/message-id/CAJvoCuuoGo5zJTpmPm90doYTUWoeUc%2BONXK2%2BH_vxsi%2BZi09bQ%40mail.gmail.com
Best Regards,
Mark Rofail
On Thu, Jul 27, 2017 at 12:54 PM, Alexander Korotkov
wrote:
>
> How many rows of FK table were referencing the PK table row you're
> updating/deleting.
> I wonder how may RI trigger work so fast if it has to do some job besides
> index search with no results?
>
The problem here is that the only to
extremely difficult to sort all of
> the problems that would result from a change in this area.
Best Regards,
Mark Rofail
>
> However, there is a bug that prevented me from testing the third scenario,
> I assume there's an issue of incompatible types problem since the right
> operand type is anyelement and the supporting procedures expect anyarray.
> I am working on debugging it right now.
>
I have also solved the bu
On Wed, Jul 19, 2017 at 10:08 PM, Alvaro Herrera
wrote:
> So let's step back a bit,
> get a patch that works for the case where the types match on both sides
> of the FK, then we review that patch; if all is well, we can discuss the
> other problem as a stretch goal.
Agreed. This should be a fu
On Wed, Jul 19, 2017 at 7:28 PM, Robert Haas wrote:
> Why do we have to solve that limitation?
Since the regress test labled element_foreing_key fails now that I made the
RI queries utilise @(anyarray, anyelement), that means it's not functioning
as it is meant to be.
ally enhances the
performance of the RI checks. How can I go about this?
Best Regards,
Mark Rofail
t4[] @>> int8,
int4[] @>> int4, int4[] @>> int2, int4[] @>> numeric.
So I think implementing operators is the way to go.
Best Regards,
Mark Rofail.
On Tue, 18 Jul 2017 at 7:43 pm, Alexander Korotkov
wrote:
> separate operators for int4[] @>> int8, int4[] @>> int4, int4[] @>> int2,
> int4[] @>> numeric.
>
My only comment on the separate operators is its high maintenance. Any new
datatype introduced a corresponding operator should be create
On Tue, 18 Jul 2017 at 7:43 pm, Alexander Korotkov
wrote:
> On T upue, Jul 18, 2017 at 2:24 AM, Mark Rofail
> wrote:
>
>> On Wed, Jul 12, 2017 at 12:53 AM, Alvaro Herrera <
>> alvhe...@2ndquadrant.com> wrote:
>>>
>>> We have one opclass for e
On Wed, Jul 12, 2017 at 12:53 AM, Alvaro Herrera
wrote:
>
> We have one opclass for each type combination -- int4 to int2, int4 to
> int4, int4 to int8, etc. You just need to add the new strategy to all
> the opclasses.
I tried this approach by manually declaring the operator multiple of time
On Wed, Jul 12, 2017 at 2:30 PM, Mark Rofail wrote:
> On Wed, Jul 12, 2017 at 12:53 AM, Alvaro Herrera > wrote:
>>
>> We have one opclass for each type combination -- int4 to int2, int4 to
>> int4, int4 to int8, etc. You just need to add the new strategy to all
>>
lution would be external
casting
BTW now that we've gone through this a little further, it's starting to
> look like a mistake to me to use the same @> operator for (anyarray,
> anyelement) than we use for (anyarray, anyarray).
I agree. Changed to @>>
Best Regards,
Mark Rofail
here are the modifications to ri_triggers.c
On Wed, Jul 12, 2017 at 12:26 AM, Mark Rofail
wrote:
>
> *What I did *
>
>- now the RI checks utilise the @>(anyarray, anyelement)
>
> Best Regards,
> Mark Rofail
>
diff --git a/src/backend/utils/adt/ri_trigger
traint
> would be violated:
SELECT 1 FROM ONLY fktable x WHERE $1 = ANY (fkcol) FOR SHARE OF x;.
or is there anything remaining ?
Best Regards,
Mark Rofail
On Sun, Jul 9, 2017 at 2:38 AM, Alexander Korotkov
wrote:
> Could you, please, specify idea of what you're implementing in more
> detail?
>
Ultimatley we would like an indexed scan instead of a sequential scan, so I
thought we needed to index the FK array columns first.
Strategy) within the CREATE INDEX query. For
example: CREATE
INDEX ON fktable USING gin (fkcolumn array_ops)
Where does the strategy number fit?
- The patch is attached here, is the approach I took to creating an
index programmatically, correct?
Best Regard,
Mark Rofail
diff
y in C index a table?
[1] https://www.postgresql.org/message-id/28389.1351094795%40sss.pgh.pa.us
Best Regards,
Mark Rofail
diff --git a/src/backend/utils/adt/ri_triggers.c b/src/backend/utils/adt/ri_triggers.c
index 3a25ba52f3..0045f64c9e 100644
--- a/src/backend/utils/adt/ri_triggers.c
+++ b/src
eck if the Datum was null and its type, I had to
do it in the arraycontainselem and pass it as a parameter to the underlying
function array_contains_elem. I'm proposing to introduce a new struct like
ArrayType, but ElementType along all with brand new MACROs to make dealing
with anyelement e
results(src/test/regress/expected/arrays.out)
- wokred on the new GIN strategy, I don't think it would vary much from
GinContainsStrategy.
*What I plan to do:*
- I need to start working on the Referential Integrity code but I don't
where to start
Best Regards,
Mark Rofail
diff --git
uch a check?
Best Regards,
Mark Rofail
diff --git a/src/backend/access/gin/ginarrayproc.c b/src/backend/access/gin/ginarrayproc.c
index cc7435e030..214aac8fba 100644
--- a/src/backend/access/gin/ginarrayproc.c
+++ b/src/backend/access/gin/ginarrayproc.c
@@ -24,6 +24,7 @@
#define GinContainsStrate
• After finding the arraycontains function, I implemented
arraycontainselem that corresponds to the operator @<(anyarray,
anyelem)
◦ Please read the attached patch file to view my progress.
• In addition to src/backend/utils/adt/arrayfuncs.c where I
implemented arraycontainselem.
◦
ed in the source code
Best Regards,
Mark Rofail
, Apr 6, 2017 at 4:18 PM, Mark Rofail wrote:
> Hello Mr Alexander,
>
> From my understanding, the main issue occurs whenever any UPDATE or
> DELETE statement is executed on the PK table,
> this triggers a referential integrity check on the FK table. In the
> previous patches
32 matches
Mail list logo
